Bull math and bear math are nearly equal and opposite — the trade is the slope, not the level.

• NIM tailwinds are mechanical. $292M of regulatory-mandated RIDF deposits earning ~3% is scheduled to redeploy into ~9% advances; CASA grew 14.9% YoY against industry 7–9%; savings rates already cut 150 bps. Each 25 bps of NIM lifts pre-tax profit by ~$105M — about 28% of FY26 net profit.
• Credit cost mean-reverts the other way. FY26 credit cost printed 0.2% — a multi-decade bottom. The same line ate $4.3B in FY20. Reversion to a mid-cycle 60–80 bps erases 80–100 bps of ROE — arithmetic equal-and-opposite of the NIM upside.
• The variant lever sits underneath. Cost-to-income fell 460 bps in a single year (71.3% → 66.7%, Q4 exit 63%) on 11.4% income growth vs 4.6% opex. This is management-controlled and largely independent of the NIM-vs-credit-cost race — and is the lever the sell-side is under-weighting.

Five clean years, one heart attack, six years of grind back to break-even.

Before: Through FY18, Yes Bank was a high-quality compounder posting 18–21% ROE on a corporate-loan growth flywheel built by founder Rana Kapoor. EPS hit $0.28 in FY18; the stock peaked at $6.20 in August 2018.

Pivot: The corporate book cracked in FY19; FY20 was the credit equivalent of a heart attack — $4.3B of provisions in one year, a $2.2B loss, ROE of −68%. The RBI imposed a Reconstruction Scheme in March 2020, wrote down $1.1B of AT-1 bonds, and brought in SBI as anchor investor. Share count expanded 6× from ~250 Cr to ~501 Cr in twelve months.

Today: The balance sheet is rebuilt — NNPA 0.2% beats every peer, CET-1 sits at 13.8%, deposits crossed $31B. SMBC closed a 20% stake in Sep 2025 — the largest cross-border investment in any Indian bank. New CEO Vinay Tonse took over April 2026. The franchise compounded back. Per-share earnings did not.

Balance sheet repaired. Earnings engine still half-built.

Pre-provision operating profit nearly tripled FY22 → FY26 ($278M → $587M, a 27% CAGR) — the underlying engine is healing. But $511M of total recoveries and upgrades (including ~$165M of SR write-backs) and a $31M write-back from one resolved corporate account are flattering FY26 net profit; strip them and core ROA falls from 0.8% to ~0.6%. The peer regression (P/B ≈ 0.13 × ROE + 0.6) prices today's earnings fairly — there is no margin of safety on the level. The position is a bet on slope.

One unique structural asset, one durable disadvantage capital cannot buy.

• SMBC is a regulatory carve-out no peer can replicate. RBI raised the single-foreign-bank holding cap from 15% to 20% specifically for the September 2025 deal; SMBC took 20% and added 4.2% from Carlyle. Two directors on the board; Moody's upgraded to Ba1 (Stable) in May 2026 citing the funding profile.
• What already works. NNPA 0.2% is the lowest in the peer set (vs ICICI 0.37%, Kotak 0.31%, HDFC 0.5%). UPI Payee PSP share at 55–57% holds #1. Non-interest income is 42% of revenue — the highest mix among Indian private banks.
• The missing piece. CASA at 35.1% sits 4–12 pp below Kotak (43%), ICICI (39%), and HDFC (38%) — a 25-year deposit-franchise gap built on millions of payroll and salary accounts that capital cannot buy. The realistic ceiling for ROE is Axis-like (13%, 3.9% NIM), not HDFC-like.

One binary tail, one earnings inflection — both land inside the next six months.

• Supreme Court AT-1 verdict — any session day. Judgment was reserved 26 Feb 2026 on the $1.1B AT-1 write-down. An adverse ruling creates a liability of ~$1.46–1.57B (principal plus six years of 9% interest), a ~16% book haircut, and a fresh equity raise diluting SMBC and existing holders 15–20%. Management's stated position: no material financial impact. The disclosed balance sheet does not provision for it.
• Q1 FY27 results — ~17 July 2026. The first print under Tonse and the first quarter where the FY26 SR-write-back cushion ($165M full-year) shrinks. Sell-side median target $0.20 from ten brokers sits below spot $0.23 — the bar is set low, and Q4 FY26 management has already pre-buffered $36M of prudent contingent provisions.
• SMBC operational follow-through — open-ended. Eight months post-close, no JV announced, no published cross-border deposit line drawn. A concrete Japan-corridor pipeline or transaction-banking JV would convert SMBC from a passive holder into an active partner — the highest-asymmetry call-side optionality in the file.